LLM API glossary
Plain-language definitions of the terms you'll meet when calling large language models through an API gateway like KeepRouter. Each entry is a short, standalone answer.
LLM API gateway
An LLM API gateway routes requests to models from different makers behind one account and key. Compatible chat models can share a client format. Specialized modalities may use dedicated endpoints.
OpenAI-compatible API
An OpenAI-compatible API accepts the OpenAI request shape, commonly POST /v1/chat/completions with an Authorization: Bearer key. Compatible clients can keep that shape after changing the base URL and API key. KeepRouter's base URL is https://keeprouter.com/v1.
Anthropic-compatible API
An Anthropic-compatible API accepts the Anthropic request shape: POST /v1/messages with x-api-key and anthropic-version headers. Claude Code and the Anthropic SDK use this endpoint. KeepRouter supports it for compatible chat models.
Token
A token is the unit a language model reads and writes, usually a word fragment or a few characters. Prompt length and output are measured in tokens. Text-model prices are usually quoted per million tokens.
Context window
A context window is the maximum number of tokens a model can consider at once, including the prompt and reply. A larger window can hold a longer document or conversation in one request.
Input, output, and cached pricing
Applicable text models can have separate input, output, and cached-input rates. KeepRouter displays the rates available for each model in the live catalog.
Prompt caching
Prompt caching reuses the processed form of a repeated prompt prefix. When a model and route support it, this can reduce the billed cached-input rate and response latency.
Per-image pricing
Image-generation models are billed per image produced instead of per token. KeepRouter displays their per-image price in the catalog.
Model maker
The model maker is the organization that trained and released a model, such as OpenAI, Anthropic, Google DeepMind, Zhipu AI, DeepSeek, ByteDance, Moonshot AI, Alibaba, or MiniMax. KeepRouter displays this metadata when it is available.
Modality
Modality describes the kinds of input and output a model handles: text models take and return text; vision (multimodal) models also accept images as input; text-to-image models generate images from a text prompt.
Streaming
Streaming returns a model's reply incrementally as it is generated instead of waiting for the whole response. On KeepRouter, it is available on compatible chat routes when the selected model and active route support it.
Tool calling
Tool calling lets a model ask your application to run a named function with structured arguments, then continue using the result. On compatible KeepRouter chat routes, supported calls pass through or are translated; availability depends on the selected model and active route.
Pay-as-you-go / prepaid credits
Pay-as-you-go means you are billed only for what you use, with no fixed monthly fee. KeepRouter uses prepaid credits: you top up a USD balance in advance and each request draws from it.
Published model pricing
KeepRouter publishes the customer price for each model in its live catalog. Text models list input, output, and any cached-input rates; image models list a per-image price. Billing is prepaid and pay-as-you-go, with no monthly subscription.
Merchant of Record
A Merchant of Record (MoR) is the company that legally sells to you and handles payment, tax, and compliance. KeepRouter's card top-ups are processed by Paddle as Merchant of Record.